Strong evidence suggests that memory for emotional information is much better than for neutral one. Thus, one may expect that forgetting of emotional information is difficult and requires considerable effort. The aim of this item-method directed forgetting functional magnetic resonance imaging study was to investigate this hypothesis both at behavioral and neural levels. In the item-method paradigm, study items are individually cued to-be-remembered or to-be-forgotten on a trial-by-trial basis: REMEMBER or FORGET instruction follows the presentation of each study item. Afterward, memory is unexpectedly tested for all items, irrespective of the previous memory instructions. Regardless of which paradigm is used, to-be-forgotten items generally show impaired recall compared with to-be-remembered items. This effect is known as the directed forgetting effect. In our study, directed forgetting effects were observed for both neutral and emotionally negative International Affective Picture System images. Moreover, recognition rate of negative to-be-forgotten images was higher than in case of neutral ones. In the study phase, intention to forget and successful forgetting of emotionally negative images were associated with widespread activations extending from the anterior to posterior regions mainly in the right hemisphere, whereas in the case of neutral images, they were associated with just one cluster of activation in the right lingual gyrus. Therefore, forgetting of emotional information seems to be a demanding process that strongly activates a distributed neural network in the right hemisphere. In the test phase, in turn, successfully forgotten images – either neutral or emotionally negative – were associated with virtually no activation, even at the lowered P value threshold. These results suggest that intentional inhibition during encoding may be an efficient strategy to cope with emotionally negative memories.